Abstract: This paper presents a new algorithm named PF-AMCMC based on Maximum Kullback-Leibler distance(abbreviated as MKLD)criterion.This algorithm can adaptively choose the numberof particles and at the same time select the implementation moment of MCMC movement,and reducesthe computational complexity under conditions guaranteeing the accuracy of state estimation.The re-sults of computational experiments and a GPS/DR integrated navigation simulation experiment showthat the improved particle filtering methods proposed in this paper have a better performances for stateestimation than other approaches.